Gnumeric · Statistical analysis · Intermediate
Create a Correlation Matrix in Gnumeric
Use the Gnumeric Correlation analysis tool to compare several measurement columns and interpret Pearson correlation.
Official documentation reviewed; native application verification pending · Updated 10 October 2026 · Editorial standards
When you want to compare two or more measurement columns, a correlation matrix gives pairwise Pearson correlation coefficients. A value near +1 shows a strong positive linear association, near −1 a negative linear association, and near zero little linear association.
A small reproducible example
Open paired variables (CSV). The file has X, Y, and Z columns. Y is deliberately calculated as 2 × X + 1.
| X | Y | Z |
|---|---|---|
| 1 | 3 | 5 |
| 2 | 5 | 4 |
| 3 | 7 | 8 |
| 4 | 9 | 7 |
The full dataset contains eight rows. The expected correlation between X and Y is 1.000, because the relationship is exactly linear.
Generate the matrix
- Import the CSV into Gnumeric and confirm all three columns contain numeric cells.
- Open the Correlation analysis dialog within the statistical tools.
- Set the input range to all three columns, including the headings if you enable Labels.
- Group the input by columns, then choose a new sheet for the result.
- Verify that the X–Y and Y–X values agree and are approximately 1.
You can cross-check the X–Y relationship with =CORREL(A2:A9,B2:B9) when those are the two imported columns. Gnumeric normally uses commas to separate function arguments in English-style locales; locales using a decimal comma may use semicolons instead. See CORREL.
Interpret results responsibly
- Correlation describes a linear association; it does not prove causation.
- Nonlinear relationships can be strong even when a Pearson coefficient is small.
- Missing data can lead to different effective sample sizes for different pairs.
- Before presenting a correlation matrix, examine a scatterplot for outliers.
Verification note: Dataset values and the X–Y relationship are independently checked. Dialog steps follow official Gnumeric documentation; native execution is pending.
Official source: Application handbook or manual. This guide is an editorial draft prepared for staging; exact menus and outputs must be checked in the target software/version before public release.